You can specify various options such as the size of resources to be allocated from the system when submitting your job.
There are two ways to do this.

1. Write the option in job script
2. Add the option when you submit the job
   

Write the appropriate option in your job script．(Refer. 1-1. Job Script)

#PBS -l <resources> 
#PBS -m be -M XXXX(at)jaist.ac.jp
#PBS -j oe           

When executing the qsub command, add the options as follows.

qsub -l <resources> <job script>
qsub -m be -M xxxxx(at)jaist.ac.jp
qsub -j oe

Resources refer to the CPU, memory size, nodes, etc. that are reserved to run the computation.
These can be specified as chunks.

A "chunk" is a group of resources, and is specified as follows.
-l select=<number>:<Resources>

Example:
  -l select=2:ncpus=8

In this example, 2 (chunk) x 8CPUs = 16 CPUs are reserved.
However, depending on the availability of resources, 8 CPUs may be allocated from each of the two physical nodes.In this case, the 8 CPUs and memory reserved for the second physical node cannot be used unless inter-node communication such as MPI.

If you want to allocate 16 CPUs from the one node, do the following.

Example2-1:
-l select=2:ncpus=8
-l place=pack

Example2-2:
-l select=1:ncpus=16

However, it may take longer time to start the job than the first example because the job will not be executed until 16 CPUs are available on a node.
Also, if you set a value higher than the number of physical CPUs in the node, the job will not run forever.
(Max CPU Num:lmpcc cluster-64CPU, KAGAYAKI CPU-Node:128CPU)

You can be notified by e-mail when a job starts or ends.

-l -m be  [b -> begin(job starts) , e -> end(job end)] 
-l -M <Mail to Address>

ジョブの終了後に，"標準出力ファイル"と"エラー出力ファイル"が生成されます．(出力内容が無い場合は空ファイルです．)
これらを一ファイルにまとめて出力させることができます．

-j oe